Doosan, Microsoft plan to build GPT-based robots

Doosan Robotics has partnered with Microsoft and Doosan Digital Innovation to give collaborative robots (cobots) a context-aware control system based on generative pre-trained transformer technology. The partnership was formalized with a memorandum of understanding for Doosan Robotics to use Microsoft’s Azure OpenAI Service for a GPT-based robot control system. 

As part of the agreement, Azure will provide Doosan with access to the GPT language model, and Doosan Digital Innovation will be responsible for adapting the model to be used in Doosan’s robots. Doosan hopes this integration will shorten programming times while also expanding the available functions for their robots. 

Doosan plans to leverage GPT’s comprehension capabilities to allow robots to gain more context and nuance in understanding tasks. GPT is a large language model that can understand and generate human-like text based on the vast amounts of data it’s trained on. 

The company hopes to integrate GPT into its system, cobots will be able to self-correct errors during operation and anticipate and adjust their actions based on the context. 

For example, Doosan’s E-SERIES cobots are designed for food and beverage preparation. With GPT integrated, a user could issue a simple voice command to prepare a dish, and the robot could figure out the optimal sequence of tasks to complete the meal. 

GPT integration also allows robots to access and learn from an existing database of code, which will make it so programmers don’t have to program every new task from scratch. 

Doosan Robotics plans to first test this technology in the food and beverage sector later this year. Depending on the outcomes of this testing, the company may explore using it in the manufacturing sector as well. 

At RoboBusiness, which takes place October 18-19 in Santa Clara, CA, a keynote panel of robotics industry leaders will discuss the applications of Large Language Models (LLMs) and text generation applications to robotics. It will also explore fundamental ways generative AI can be applied to robotics design, model training, simulation, control algorithms and product commercialization.

The panel will include Pras Velagapudi, VP of Innovation at Agility Robotics, Jeff Linnell, CEO and Founder of Formant, Ken Goldberg, the William S. Floyd Jr. Distinguished Chair in Engineering at UC Berkeley, Amit Goel, the Director of Product Management at NVIDIA, and Ted Larson, the CEO of OLogic.